feat: custom artist separator regex

docs/src/content/docs/reference/configuration.md

@@ -40,6 +40,9 @@ If the environment variable is defined without **and** with the suffix at the sa
 ##### KOITO_LOG_LEVEL
 - Default: `info`
 - Description: One of `debug | info | warn | error | fatal`
+##### KOITO_ARTIST_SEPARATORS_REGEX
+- Default: `\s+·\s+`
+- Description: The list of regex patterns Koito will use to separate artist strings, separated by two semicolons (`;;`). 
 ##### KOITO_MUSICBRAINZ_URL
 - Default: `https://musicbrainz.org`
 - Description: The URL Koito will use to contact MusicBrainz. Replace this value if you have your own MusicBrainz mirror.

internal/catalog/associate_artists.go

@@ -62,7 +62,7 @@ func AssociateArtists(ctx context.Context, d db.DB, opts AssociateArtistsOpts) (
 	}
 
 	if len(result) < 1 {
-		allArtists := slices.Concat(opts.ArtistNames, ParseArtists(opts.ArtistName, opts.TrackTitle))
+		allArtists := slices.Concat(opts.ArtistNames, ParseArtists(opts.ArtistName, opts.TrackTitle, cfg.ArtistSeparators()))
 		l.Debug().Msgf("Associating artists by artist name(s) %v and track title '%s'", allArtists, opts.TrackTitle)
 		fallbackMatches, err := matchArtistsByNames(ctx, allArtists, nil, d, opts)
 		if err != nil {
@@ -180,7 +180,7 @@ func matchArtistsByMBID(ctx context.Context, d db.DB, opts AssociateArtistsOpts,
 		}
 
 		if len(opts.ArtistNames) < 1 {
-			opts.ArtistNames = slices.Concat(opts.ArtistNames, ParseArtists(opts.ArtistName, opts.TrackTitle))
+			opts.ArtistNames = slices.Concat(opts.ArtistNames, ParseArtists(opts.ArtistName, opts.TrackTitle, cfg.ArtistSeparators()))
 		}
 
 		a, err = resolveAliasOrCreateArtist(ctx, id, opts.ArtistNames, d, opts)

internal/catalog/catalog.go

@@ -201,21 +201,18 @@ func buildArtistStr(artists []*models.Artist) string {
 var (
 	// Bracketed feat patterns
 	bracketFeatPatterns = []*regexp.Regexp{
-		regexp.MustCompile(`(?i)\(feat\. ([^)]*)\)`),
+		regexp.MustCompile(`(?i)\([fF]eat\. ([^)]*)\)`),
-		regexp.MustCompile(`(?i)\[feat\. ([^\]]*)\]`),
+		regexp.MustCompile(`(?i)\[[fF]eat\. ([^\]]*)\]`),
 	}
 	// Inline feat (not in brackets)
-	inlineFeatPattern = regexp.MustCompile(`(?i)feat\. ([^()\[\]]+)$`)
+	inlineFeatPattern = regexp.MustCompile(`(?i)[fF]eat\. ([^()\[\]]+)$`)
 
 	// Delimiters only used inside feat. sections
 	featSplitDelimiters = regexp.MustCompile(`(?i)\s*(?:,|&|and|·)\s*`)
-
-	// Delimiter for separating artists in main string (rare but real usage)
-	mainArtistDotSplitter = regexp.MustCompile(`\s+·\s+`)
 )
 
 // ParseArtists extracts all contributing artist names from the artist and title strings
-func ParseArtists(artist string, title string) []string {
+func ParseArtists(artist string, title string, addlSeparators []*regexp.Regexp) []string {
 	seen := make(map[string]struct{})
 	var out []string
 
@@ -230,12 +227,9 @@ func ParseArtists(artist string, title string) []string {
 		}
 	}
 
-	foundFeat := false
-
 	// Extract bracketed features from artist
 	for _, re := range bracketFeatPatterns {
 		if matches := re.FindStringSubmatch(artist); matches != nil {
-			foundFeat = true
 			artist = strings.Replace(artist, matches[0], "", 1)
 			for _, name := range featSplitDelimiters.Split(matches[1], -1) {
 				add(name)
@@ -244,7 +238,6 @@ func ParseArtists(artist string, title string) []string {
 	}
 	// Extract inline feat. from artist
 	if matches := inlineFeatPattern.FindStringSubmatch(artist); matches != nil {
-		foundFeat = true
 		artist = strings.Replace(artist, matches[0], "", 1)
 		for _, name := range featSplitDelimiters.Split(matches[1], -1) {
 			add(name)
@@ -252,14 +245,19 @@ func ParseArtists(artist string, title string) []string {
 	}
 
 	// Add base artist(s)
-	if foundFeat {
+	l1 := len(out)
-		add(strings.TrimSpace(artist))
+	for _, re := range addlSeparators {
-	} else {
+		for _, name := range re.Split(artist, -1) {
-		// Only split on " · " in base artist string
+			if name == artist {
-		for _, name := range mainArtistDotSplitter.Split(artist, -1) {
+				continue
+			}
 			add(name)
 		}
 	}
+	// Only add the full artist string if no splitters were matched
+	if l1 == len(out) {
+		add(artist)
+	}
 
 	// Extract features from title
 	for _, re := range bracketFeatPatterns {

internal/catalog/catalog_test.go

@@ -5,6 +5,7 @@ import (
 	"fmt"
 	"log"
 	"os"
+	"regexp"
 	"testing"
 	"time"
 
@@ -167,15 +168,15 @@ func getTestGetenv(resource *dockertest.Resource) func(string) string {
 
 func truncateTestData(t *testing.T) {
 	err := store.Exec(context.Background(),
-		`TRUNCATE 
+		`TRUNCATE
-		artists, 
+		artists,
 		artist_aliases,
-		tracks, 
+		tracks,
-		artist_tracks, 
+		artist_tracks,
-		releases, 
+		releases,
-		artist_releases, 
+		artist_releases,
 		release_aliases,
-		listens 
+		listens
 		RESTART IDENTITY CASCADE`)
 	require.NoError(t, err)
 }
@@ -184,23 +185,23 @@ func setupTestDataWithMbzIDs(t *testing.T) {
 	truncateTestData(t)
 
 	err := store.Exec(context.Background(),
-		`INSERT INTO artists (musicbrainz_id) 
+		`INSERT INTO artists (musicbrainz_id)
 			VALUES ('00000000-0000-0000-0000-000000000001')`)
 	require.NoError(t, err)
 	err = store.Exec(context.Background(),
-		`INSERT INTO artist_aliases (artist_id, alias, source, is_primary) 
+		`INSERT INTO artist_aliases (artist_id, alias, source, is_primary)
 			VALUES (1, 'ATARASHII GAKKO!', 'Testing', true)`)
 	require.NoError(t, err)
 	err = store.Exec(context.Background(),
-		`INSERT INTO releases (musicbrainz_id) 
+		`INSERT INTO releases (musicbrainz_id)
 			VALUES ('00000000-0000-0000-0000-000000000101')`)
 	require.NoError(t, err)
 	err = store.Exec(context.Background(),
-		`INSERT INTO release_aliases (release_id, alias, source, is_primary) 
+		`INSERT INTO release_aliases (release_id, alias, source, is_primary)
 			VALUES (1, 'AG! Calling', 'Testing', true)`)
 	require.NoError(t, err)
 	err = store.Exec(context.Background(),
-		`INSERT INTO artist_releases (artist_id, release_id) 
+		`INSERT INTO artist_releases (artist_id, release_id)
 			VALUES (1, 1)`)
 	require.NoError(t, err)
 	err = store.Exec(context.Background(),
@@ -221,23 +222,23 @@ func setupTestDataSansMbzIDs(t *testing.T) {
 	truncateTestData(t)
 
 	err := store.Exec(context.Background(),
-		`INSERT INTO artists (musicbrainz_id) 
+		`INSERT INTO artists (musicbrainz_id)
 			VALUES (NULL)`)
 	require.NoError(t, err)
 	err = store.Exec(context.Background(),
-		`INSERT INTO artist_aliases (artist_id, alias, source, is_primary) 
+		`INSERT INTO artist_aliases (artist_id, alias, source, is_primary)
 			VALUES (1, 'ATARASHII GAKKO!', 'Testing', true)`)
 	require.NoError(t, err)
 	err = store.Exec(context.Background(),
-		`INSERT INTO releases (musicbrainz_id) 
+		`INSERT INTO releases (musicbrainz_id)
 			VALUES (NULL)`)
 	require.NoError(t, err)
 	err = store.Exec(context.Background(),
-		`INSERT INTO release_aliases (release_id, alias, source, is_primary) 
+		`INSERT INTO release_aliases (release_id, alias, source, is_primary)
 			VALUES (1, 'AG! Calling', 'Testing', true)`)
 	require.NoError(t, err)
 	err = store.Exec(context.Background(),
-		`INSERT INTO artist_releases (artist_id, release_id) 
+		`INSERT INTO artist_releases (artist_id, release_id)
 			VALUES (1, 1)`)
 	require.NoError(t, err)
 	err = store.Exec(context.Background(),
@@ -358,10 +359,16 @@ func TestArtistStringParse(t *testing.T) {
 		// artists in both
 		{"Daft Punk feat. Julian Casablancas", "Instant Crush (feat. Julian Casablancas)"}:   {"Daft Punk", "Julian Casablancas"},
 		{"Paramore (feat. Joy Williams)", "Hate to See Your Heart Break feat. Joy Williams"}: {"Paramore", "Joy Williams"},
+		{"MINSU", "오해 금지 (Feat. BIG Naughty)"}:                                               {"MINSU", "BIG Naughty"},
+		{"MINSU", "오해 금지 [Feat. BIG Naughty]"}:                                               {"MINSU", "BIG Naughty"},
+		{"MINSU", "오해 금지 Feat. BIG Naughty"}:                                                 {"MINSU", "BIG Naughty"},
+
+		// custom separator
+		{"MIMiNARI//楠木ともり", "眠れない"}: {"MIMiNARI", "楠木ともり"},
 	}
 
 	for in, out := range cases {
-		artists := catalog.ParseArtists(in.Name, in.Title)
+		artists := catalog.ParseArtists(in.Name, in.Title, []*regexp.Regexp{regexp.MustCompile(`\s*//\s*`), regexp.MustCompile(`\s+·\s+`)})
 		assert.ElementsMatch(t, out, artists)
 	}
 }

internal/cfg/cfg.go

@@ -3,6 +3,7 @@ package cfg
 import (
 	"errors"
 	"fmt"
+	"regexp"
 	"strconv"
 	"strings"
 	"sync"
@@ -45,6 +46,7 @@ const (
 	IMPORT_BEFORE_UNIX_ENV         = "KOITO_IMPORT_BEFORE_UNIX"
 	IMPORT_AFTER_UNIX_ENV          = "KOITO_IMPORT_AFTER_UNIX"
 	FETCH_IMAGES_DURING_IMPORT_ENV = "KOITO_FETCH_IMAGES_DURING_IMPORT"
+	ARTIST_SEPARATORS_ENV          = "KOITO_ARTIST_SEPARATORS_REGEX"
 )
 
 type config struct {
@@ -80,6 +82,7 @@ type config struct {
 	userAgent              string
 	importBefore           time.Time
 	importAfter            time.Time
+	artistSeparators       []*regexp.Regexp
 }
 
 var (
@@ -189,6 +192,18 @@ func loadConfig(getenv func(string) string, version string) (*config, error) {
 	rawCors := getenv(CORS_ORIGINS_ENV)
 	cfg.allowedOrigins = strings.Split(rawCors, ",")
 
+	if getenv(ARTIST_SEPARATORS_ENV) != "" {
+		for pattern := range strings.SplitSeq(getenv(ARTIST_SEPARATORS_ENV), ";;") {
+			regex, err := regexp.Compile(pattern)
+			if err != nil {
+				return nil, fmt.Errorf("failed to compile regex pattern %s", pattern)
+			}
+			cfg.artistSeparators = append(cfg.artistSeparators, regex)
+		}
+	} else {
+		cfg.artistSeparators = []*regexp.Regexp{regexp.MustCompile(`\s+·\s+`)}
+	}
+
 	switch strings.ToLower(getenv(LOG_LEVEL_ENV)) {
 	case "debug":
 		cfg.logLevel = 0
@@ -388,3 +403,9 @@ func FetchImagesDuringImport() bool {
 	defer lock.RUnlock()
 	return globalConfig.fetchImageDuringImport
 }
+
+func ArtistSeparators() []*regexp.Regexp {
+	lock.RLock()
+	defer lock.RUnlock()
+	return globalConfig.artistSeparators
+}